Teamtailor

Name: Teamtailor MCP Server
Availability: InStock
Author: Teamtailor

Productivity45 tools

Query jobs, candidates and hiring pipelines, move applications between stages and manage todos from your AI assistant. 45 tools, OAuth sign-in, and the first connector in this catalogue where delete is its own separately grantable scope.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on 18 August 2026.

Anthropic states this reflects the level of review a connector received, not a security audit.

Connect Teamtailor via MCP

https://mcp.teamtailor.com/mcp

Works in any MCP-compatible client. In Agentman Studio it is one click — no config file to edit.

Teamtailor Tools & Capabilities (45)

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

It reads other people's job applications. Candidates are third parties who applied to a company, not to an AI assistant. Teamtailor's MCP Privacy Policy lists names, emails, phone numbers, locations, LinkedIn URLs, résumé text and file links among what the tools return, and warns that free-text fields, messages, transcripts and uploaded documents may contain sensitive personal data. That material lands in your chat transcript, subject to your AI provider's retention rules rather than Teamtailor's.
Teamtailor's no-training promise covers Teamtailor only. Its policy says twice that MCP data is not used to train Teamtailor's models, and immediately notes that the connected client provider's terms govern its own use, including training. Read those before connecting.
move_application_to_stage advances a real person's application and needs only write access. It sits under Create & modify, not Delete, and Teamtailor documents no MCP-level confirmation step before it runs. Whether your client asks for approval is a property of the client.
Blind hiring through MCP is a design statement, not a measurement. Teamtailor says the server is designed to apply Anonymous mode masking. We never authenticated, so we did not observe it.
Audit logs are short by default. Fifteen days, extendable to 30 in settings, and longer only at additional cost. The audit log is itself an add-on a Company Admin must activate. If nobody turned it on, there is no MCP trail to read.
Permission changes may not take effect until you reconnect. Teamtailor states you may need to re-authenticate or refresh the connection after changing MCP permissions, and that the steps depend on the client. Narrowing a role's access is not reliably immediate.
Google Gemini cannot connect. Teamtailor states Gemini supports only officially partnered MCP servers, so third-party servers including this one cannot be added.
Compliance is explicitly the customer's job. Teamtailor states the customer is responsible for determining which AI clients may connect and for ensuring its use complies with the EU AI Act and other applicable law, including measures for human oversight and review of AI-generated outputs. Teamtailor says it does not have sufficient insight into your configuration to assess this.
The scope-to-tool mapping is Teamtailor's claim, not our observation. We read the scope list from the server's own descriptors, but the assignment of each tool to a scope comes from Teamtailor's documentation. We read no per-tool annotations, because the handshake is gated.
We did not exercise any tool. Our check was an anonymous protocol request that returned an OAuth challenge, plus reads of the server's two .well-known descriptors. Everything about tool behaviour comes from Teamtailor's documentation or the directory listing.

Frequently asked questions

How do you connect Teamtailor to Claude?

A Company Admin activates MCP in the Add-on feature center first. Teamtailor then documents the Claude path as Organization settings, Customize, Connectors, Add, Browse connectors, finding the Teamtailor connector and clicking Add. You sign in with Teamtailor, choose which account to connect if you have several, and click Authorize.

Does the Teamtailor MCP server cost extra?

No. Teamtailor lists MCP in its Add-on feature center at no additional cost, alongside features such as interview kits and job offers. A Company Admin still has to activate it before anyone can connect, and the selected Teamtailor account must have the add-on enabled or the sign-in step will not complete.

Can you give an AI assistant read and write access to Teamtailor without letting it delete?

Yes. Teamtailor's MCP settings let a Company Admin set each user role to read, create and modify, or delete access, and the server advertises read, write and delete as three separate OAuth scopes. Delete is the only level that reaches the six removal tools, so it can be withheld while writing stays available.

Which Teamtailor tools can delete data?

Six, and Teamtailor names them under its own Delete heading. They are archive_question, remove_job_question, remove_job_interview_kit, remove_job_skill_or_trait, remove_tag and delete_todo. None of them delete a candidate, an application or a job. The connector's destructive surface is job configuration, tags and todos only.

Can the Teamtailor MCP server see candidate data the connected user cannot?

No. Teamtailor states that AI assistants never receive their own permissions and connect through an individual Teamtailor user account. Every action is limited by both that user's existing Teamtailor permissions and the MCP access level a Company Admin set for their role, so the stricter of the two always applies.

Does connecting Teamtailor to an AI assistant show up in the audit log?

Yes. Teamtailor's audit log records a distinct actor type called User via MCP for actions taken by a connected AI assistant, alongside System, Candidate and User. Logs are accessible for 15 days by default, extendable to 30 in settings, with longer retention available at extra cost.

Does Teamtailor train AI models on data reached through the MCP server?

No. Teamtailor's dedicated MCP Server Privacy Policy states twice that data accessed through the MCP server is not used by Teamtailor to train AI models. It also warns that the statement covers Teamtailor only, and that the connected assistant's provider governs its own use of data, including training.

Are anonymized Teamtailor candidates still anonymous through MCP?

Teamtailor says the server is designed to apply the same masking. Its MCP Privacy Policy states that where a candidate is anonymized under the blind hiring feature, name, email, phone, location, LinkedIn URL and file names are withheld and identifying details are scrubbed from free-text fields. We did not test this behaviour.